This paper presents a new computer-facilitated approach for incident management to improve typical incident management. Our approach automates typical manual-based incident resolution process by proposing a new semantic similarity measurement between a given incident call and incident cases stored already in a case base. The proposed semantic similarity measurement distinguishes traditional similarity measures by incorporating additional useful information and exploiting semantic knowledge about features appeared in two incident descriptions to be compared. First, we state how typical incident management is processed and what its problems are. We then propose our automated incident resolution process with its core components. After that we introduce our identified additional useful information for our similarity measurement and describe how our similarity measurement algorithm is carried out. In an experimental evaluation, we show the technical coherence and feasibility of the proposed solution using a real dataset.